Initializing a struct field that is a vector in Julia - julia

I have the following mutable struct:
mutable struct foo
x::Vector{String}
# other field(s)...
# constructor(s)
end
I would like to create an object from this struct and edit it like the following:
bar = foo() # or some other constructor
push!(bar.x, "a")
# Do some stuff...
push!(bar.x, "b")
# Do some more stuff...
push!(bar.x, "c")
To do this, what is the best constructor for foo?
P.S. I have tried the following constructors:
foo() = new()
If I use this and do push!(bar.x, "a"), I get an UndefRefError. I could initialize bar.x outside (like bar.x = []), but I really want to avoid this.
I have also tried:
foo() = new([])
With this, I can do the push operations without a problem. However, if I have many other other fields in the struct that are also vectors, I would have to do this:
mutable struct foo
x::Vector{String}
y::Vector{String}
z::Vector{String}
w::Vector{String}
# maybe dozens more...
foo() = new([], [], [], [], ...) # kind of ugly and wastes time to type
end
Is this the best there is?

You can do:
julia> mutable struct Foo
x::Vector{String}
Foo() = new(String[])
end
julia> Foo()
Foo(String[])
However, for your scenario the most convenient would be Base.#kwdef:
Base.#kwdef mutable struct FooB
x::Vector{String} = String[]
y::Vector{String} = String[]
end
Now of course you can do:
julia> FooB()
FooB(String[], String[])
However other methods are available too:
julia> methods(FooB)
# 3 methods for type constructor:
[1] FooB(; x, y) in Main at util.jl:478
[2] FooB(x::Vector{String}, y::Vector{String}) in Main at REPL[9]:2
[3] FooB(x, y) in Main at REPL[9]:2
So you could do:
julia> FooB(;y=["hello","world"])
FooB(String[], ["hello", "world"])

There's one alternative that is quite concise and almost as fast as writing out the inputs:
struct Foo # Typenames should be uppercase
x::Vector{String}
y::Vector{String}
z::Vector{String}
w::Vector{String}
Foo() = new((String[] for _ in 1:4)...) # using a generator and splatting
# Foo() = new(ntuple(_->String[], 4)...) # also possible
end

Related

How to dispatch based on the type of any of the splatted args?

Consider an existing function in Base, which takes in a variable number of arguments of some abstract type T. I have defined a subtype S<:T and would like to write a method which dispatches if any of the arguments is my subtype S.
As an example, consider function Base.cat, with T being an AbstractArray and S being some MyCustomArray <: AbstractArray.
Desired behaviour:
julia> v = [1, 2, 3];
julia> cat(v, v, v, dims=2)
3×3 Array{Int64,2}:
1 1 1
2 2 2
3 3 3
julia> w = MyCustomArray([1,2,3])
julia> cat(v, v, w, dims=2)
"do something fancy"
Attempt:
function Base.cat(w::MyCustomArray, a::AbstractArray...; dims)
pritnln("do something fancy")
end
But this only works if the first argument is MyCustomArray.
What is an elegant way of achieving this?
I would say that it is not possible to do it cleanly without type piracy (but if it is possible I would also like to learn how).
For example consider cat that you asked about. It has one very general signature in Base (actually not requiring A to be AbstractArray as you write):
julia> methods(cat)
# 1 method for generic function "cat":
[1] cat(A...; dims) in Base at abstractarray.jl:1654
You could write a specific method:
Base.cat(A::AbstractArray...; dims) = ...
and check if any of elements of A is your special array, but this would be type piracy.
Now the problem is that you cannot even write Union{S, T} as since S <: T it will be resolved as just T.
This would mean that you would have to use S explicitly in the signature, but then even:
f(::S, ::T) = ...
f(::T, ::S) = ...
is problematic and a compiler will ask you to define f(::S, ::S) as the above definitions lead to dispatch ambiguity. So, even if you wanted to limit the number of varargs to some maximum number you would have to annotate types for all divisions of A into subsets to avoid dispatch ambiguity (which is doable using macros, but grows the number of required methods exponentially).
For general usage, I concur with Bogumił, but let me make an additional comment. If you have control over how cat is called, you can at least write some kind of trait-dispatch code:
struct MyCustomArray{T, N} <: AbstractArray{T, N}
x::Array{T, N}
end
HasCustom() = Val(false)
HasCustom(::MyCustomArray, rest...) = Val(true)
HasCustom(::AbstractArray, rest...) = HasCustom(rest...)
# `IsCustom` or something would be more elegant, but `Val` is quicker for now
Base.cat(::Val{true}, args...; dims) = println("something fancy")
Base.cat(::Val{false}, args...; dims) = cat(args...; dims=dims)
And the compiler is cool enough to optimize that away:
julia> args = (v, v, w);
julia> #code_warntype cat(HasCustom(args...), args...; dims=2);
Variables
#self#::Core.Compiler.Const(cat, false)
#unused#::Core.Compiler.Const(Val{true}(), false)
args::Tuple{Array{Int64,1},Array{Int64,1},MyCustomArray{Int64,1}}
Body::Nothing
1 ─ %1 = Main.println("something fancy")::Core.Compiler.Const(nothing, false)
└── return %1
If you don't have control over calls to cat, the only resort I can think of to make the above technique work is to overdub methods containing such call, to replace matching calls by the custom implementation. In which case you don't even need to overload cat, but can directly replace it by some mycat doing your fancy stuff.

Invoking parent functors in Julia

Since version 1.3 Julia allows functor dispatch on abstract types. Therefore, I was wondering whether it is possible to explicitly invoke the parent functor from a child object.
E.g. in the example below, is there any way to call (x::Foo)() through the bar::Bar object?
abstract type Foo end
(x::Foo)() = "Invoked Foo functor."
struct Bar <: Foo end
(x::Bar)() = "Invoked Bar functor."
bar = Bar()
#info "Calling bar functor directly."
bar() |> println
#info "Invoking parent functor."
# ??? - how to invoke parent functor (x::Foo)() (e.g. desired output "Invoked Foo functor")
invoke(bar,Tuple{Bar}, bar) |> println
what about using a default struct?
abstract type Foo end
(x::Foo)() = "Invoked Foo functor."
struct Bar <: Foo end
(x::Bar)() = "Invoked Bar functor."
struct DefaultFoo <: Foo end
#here we dont define an specialized method, DefaultFoo calls Foo
#an interface to define default methods:
default(x::Type{Foo}) = DefaultFoo
function parent(x::T) where {T}
y = default(supertype(T))
return y()
end
finally, you can do this to call the default function:
bar = Bar()
foo = parent(bar)
foo()
this requires a definition of a defaultFoo type and a default(x::Type{T}) for each supertype . you can automate this with the following macro:
macro create_default_functor(type)
a = gensym(type)
esc(quote
struct $a <: $type end
default(x::Type{$type}) = $a
end)
end
using the macro, and your code:
abstract type Foo end
(x::Foo)() = "Invoked Foo functor."
#create_default_functor Foo
struct Bar <: Foo end
(x::Bar)() = "Invoked Bar functor."
function parent(x::T) where {T}
y = default(supertype(T))
return y()
end
#calling bar
bar = Bar()
bar()
#calling foo
foo = parent(bar)
foo()
I don't have the macro knowledge right now to call the macro directly on the abstract type definition, but is a start.
The thing about abstract functors is that is a very new feature (1.3 is not out yet) and maybe this can be added on future versions of julia (something like call_parent(Foo,args...), ) if you add a PR suggesting the feature.

custom ordering in Julia SortedSet

In the Julia documentation for SortedSet, there is a reference to "ordering objects", which can be used in the constructor. I'm working on a project where I need to implement a custom sort on a set of structs. I'd like to use a functor for this, since there is additional state I need for my comparisons.
Here is a somewhat simplified version of the problem I want to solve. I have two structs, Point and Edge:
struct Point{T<:Real}
x::T
y::T
end
struct Edge{T<:Real}
first::Point{T}
second::Point{T}
end
I have a Point called 'vantage', and I want to order Edges by their distance from 'vantage'. Conceptually:
function edge_ordering(vantage::Point, e1::Edge, e2::Edge)
d1 = distance(vantage, e1)
d2 = distance(vantage, e2)
return d1 < d2
end
Are "ordering objects" functors (or functor-ish)? Is there some other conventional way of doing this sort of ordering in Julia?
An Ordering object can contain fields, you can store your state there. This is an example of a Remainder Ordering which sort integers by it's remainder:
using DataStructures
struct RemainderOrdering <: Base.Order.Ordering
r::Int
end
import Base.Order.lt
lt(o::RemainderOrdering, a, b) = isless(a % o.r, b % o.r)
SortedSet(RemainderOrdering(3), [1,2,3]) # 3, 1, 2
I'm not sure how it is related to functors, so I may misunderstand your question. This is an alternative implementation that defines an Ordering functor. I made explanations in comments.
using DataStructures
import Base: isless, map
struct Foo # this is your structure
x::Int
end
struct PrimaryOrdered{T, F} # this is the functor, F is the additional state.
x::T
end
map(f::Base.Callable, x::T) where {T <: PrimaryOrdered} = T(f(x.x)) # this makes it a functor?
isless(x::PrimaryOrdered{T, F}, y::PrimaryOrdered{T, F}) where {T, F} =
F(x.x) < F(y.x) # do comparison with your additional state, here I assume it is a closure
const OrderR3 = PrimaryOrdered{Foo, x -> x.x % 3} # a order that order by the remainder by 3
a = OrderR3(Foo(2))
f(x::Foo) = Foo(x.x + 1) # this is a Foo -> Foo
a = map(f, a) # you can map f on a OrderR3 object
a == OrderR3(Foo(33)) # true
a = map(OrderR3 ∘ Foo, [1, 2, 3])
s = SortedSet(a)
map(x->x.x, s) # Foo[3, 1, 2]
As always, an MWE is important for a question to be understood better. You can include a piece of code to show how you want to construct and use your SortedSet, instead of the vague "state" and "functor".
The sorting is based on the method isless for the type. So for instance if you have a type in which you want to sort on the b field. For instance you can do
struct Foo{T}
a::T
b::T
end
Base.:isless(x::T,y::T) where {T<:Foo} = isless(x.b,y.b)
s=[Foo(1,2),Foo(2,-1)]
res=SortedSet(s)
#SortedSet(Foo[Foo(2, -1), Foo(1, 2)],
#Base.Order.ForwardOrdering())
Tuples are also sorted in order, so you can also use
sort(s,by=x->(x.b,x.a)) to sort by b,thena without having to define isless for the type.

Julia: Best practice to unpack parameters inside a function

I can unpack a tuple. I'm trying to write a function (or macro) that would unpack a subset of these from an instance of the type-constructor Parameters(). That is, I know how to do:
a,b,c = unpack(p::Parameters)
But I would like to do something like this:
b,c = unpack(p::Parameters, b,c)
or maybe even lazier:
unpack(p::Parameters, b, c)
This is to avoid writing things like:
function unpack_all_oldstyle(p::Parameters)
a=p.a; b=p.b; c=p.c; ... z=p.z;
return a,b,c,...,z
end
There's something wrong with my approach, but hopefully there is a fix.
In case it wasn't clear from the wording of my question, I'm a total ignoramus. I read about unpacking the ellipsis here: how-to-pass-tuple-as-function-arguments
"module UP tests Unpacking Parameters"
module UP
struct Parameters
a::Int64
b::Int64
c::Int64
end
"this method sets default parameters and returns a tuple of default values"
function Parameters(;
a::Int64 = 3,
b::Int64 = 11,
c::Int64 = 101
)
Parameters(a, b, c)
end
"this function unpacks all parameters"
function unpack_all(p::Parameters)
return p.a, p.b, p.c
end
"this function tests the unpacking function: in the body of the function one can now refer to a rather than p.a : worth the effort if you have dozens of parameters and complicated expressions to compute, e.g. type (-b+sqrt(b^2-4*a*c))/2/a instead of (-p.b+sqrt(p.b^2-4*p.a *p.c))/2/p.a"
function unpack_all_test(p::Parameters)
a, b, c = unpack_all(p)
return a, b, c
end
"""
This function is intended to unpack selected parameters. The first, unnamed argument is the constructor for all parameters. The second argument is a tuple of selected parameters.
"""
function unpack_selected(p::Parameters; x...)
return p.x
end
function unpack_selected_test(p::Parameters; x...)
x = unpack_selected(p, x)
return x
end
export Parameters, unpack_all, unpack_all_test, unpack_selected, unpack_selected_test
end
p = UP.Parameters() # make an instance
UP.unpack_all_test(p)
## (3,11,101) ## Test successful
UP.unpack_selected_test(p, 12)
## 12 ## intended outcome
UP.unpack_selected_test(p, b)
## 11 ## intended outcome
UP.unpack_selected_test(p, c, b, a)
## (101,11,3) ## intended outcome
There already exists one: Parameters.jl.
julia> using Parameters
julia> struct Params
a::Int64
b::Int64
c::Int64
end
julia> #unpack a, c = Params(1,2,3)
Params(1,2,3)
julia> a,c
(1,3)
julia> #with_kw struct Params
a::Int64 = 3
b::Int64 = 11
c::Int64 = 101
end
julia> #unpack c,b,a = Params()
Params
a: Int64 3
b: Int64 11
c: Int64 101
julia> c,b,a
(101,11,3)
BTW, you can fix your unpack_selected by:
unpack_selected(p::Parameters, fields...) = map(x->getfield(p, x), fields).
# note that, the selected field names should be Symbol here
julia> unpack_selected(p, :b)
(11,)
julia> unpack_selected(p, :c, :b, :a)
(101,11,3)

How to make type attribute be a function of other type attributes?

To declare a new composite type, we use the following syntax
type foo
a::Int64
b::Int64
end
and instantiate like such
x = foo(1,3)
Is there some way to have type attributes that always just a function of other attributes? For example, is there some way to do the following (which is invalid syntax)...
type foo
a::Int64
b::Int64
c = a + b
end
My current workaround is just to define a function which calculates c and returns an instance of the type, like so...
type foo
a::Int64
b::Int64
c::Int64
end
function foo_maker(a, b)
return foo(a, b, a+b)
end
Is there a more elegant solution? Possibly one that can be contained within the type definition?
EDIT - 3/7/14
With Cristóvão's suggestion in mind, I've ended up declaring constructors like the following to allow for keyword args and attributes calculated upon instantiation
# Type with optional keyword argument structure
type LargeType
# Declare all the attributes in order up top
q::Int64
w::Int64
e::Int64
r::Int64
t::Int64
y::Int64
a::Number
b::Number
c::Number
# Declare Longer constructor with stuff going on in the body
LargeType(;q=1,w=1,e=1,r=1,t=1,y=1) = begin
# Large Constructor Example
a = round(r^t - log(pi))
b = a % t
c = a*b
# Return new instance with correctly ordered arguments
return new(q,w,e,r,t,y,a,b,c)
end
end
println(LargeType(r=2,t=5))
Try this:
julia> type foo
a::Int64
b::Int64
c::Int64
foo(a::Int64, b::Int64) = new(a, b, a+b)
end
julia> foo(1,2)
foo(1,2,3)
julia> foo(4,5,6)
no method foo(Int64, Int64, Int64)
However, that won't prevent one from manually changing a, b or c and rendering c inconsistent. To prevent that, if it presents no other problems, you can make foo immutable:
julia> immutable foo
...
There isn't any way to do this currently, but there might be in the future:
https://github.com/JuliaLang/julia/issues/1974

Resources